Dehatebert Mono Arabic

🚀 Arabic Monolingual Hate Speech Detection Model

This model is designed to detect hate speech in the Arabic language. The "mono" in its name indicates a monolingual setting, where the model is trained solely on Arabic language data. It is fine - tuned on a multilingual BERT model.

🚀 Quick Start

This model is used for detecting hate speech in Arabic. It is based on a multilingual BERT model and fine - tuned with Arabic data.

✨ Features

• Monolingual Focus: Trained specifically on Arabic language data for better performance in Arabic hate speech detection.
• Fine - Tuned on BERT: Leverages the pre - trained knowledge of a multilingual BERT model.
• Optimized Learning Rate: Trained with different learning rates, and the best validation score of 0.877609 was achieved at a learning rate of 2e - 5.

📚 Documentation

The model is trained with various learning rates, and the best validation score of 0.877609 was obtained at a learning rate of 2e - 5. You can find the training code at this url.

📄 License

This project is licensed under the Apache 2.0 license.

📖 Citation

For more details about our paper

Sai Saketh Aluru, Binny Mathew, Punyajoy Saha and Animesh Mukherjee. "Deep Learning Models for Multilingual Hate Speech Detection". Accepted at ECML - PKDD 2020.

Please cite our paper in any published work that uses any of these resources.

@article{aluru2020deep,
  title={Deep Learning Models for Multilingual Hate Speech Detection},
  author={Aluru, Sai Saket and Mathew, Binny and Saha, Punyajoy and Mukherjee, Animesh},
  journal={arXiv preprint arXiv:2004.06465},
  year={2020}
}